Purpose: Electrical stimulation mapping (ESM) is the gold standard for identifying language-relevant cortex prior to neurosurgical resections near the eloquent cortex. However, its application mode is not standardized, as some procedures rely on a single task, whereas others depend on task batteries. In this study, we examine whether multitask ESM is necessary, quantify the information loss that occurs from restricting the number of applied tasks, and search for a procedure that allows for time-efficient, multitask ESM.
Methods: Thirty-eight epilepsy and 11 brain tumor patients with subdural grid electrodes receive extraoperative ESM for language cortex at 1,890 stimulation sites. The applied test battery includes counting, naming, repeating, reading, Token Test, and body commands. Correlation analysis and principal component analysis are used to identify the functional overlap between tasks. Sensitivity analysis is the basis for determining the optimal task order, which requires a minimum number of stimulations for identifying language loci.
Results: Correlation coefficients between tasks are -0.17 to 0.63; frontal and temporoparietal language cortex are organized differently. Naming is the most sensitive task, but 31% (temporoparietal) to 43% (frontal) of language sites remained undetected after naming alone. The optimal procedure starts with naming and the Token Test and differs thereafter between frontal and temporoparietal cortex. Omitting the Token Test and body commands from the battery results in up to 26% information loss.
Discussion: Despite considerable overlap between different tasks, multitask ESM appears necessary to avoid missing language relevant cortex. Applying the tasks in an optimal order allows economizing the procedure.